Daniel Resio: BS/MS in Ocean Engineering from Florida Atlantic University, specializing in autonomous marine vehicles, control optimization, and machine/reinforcment learning. Currently building open-source robotics and AI tooling, from physics simulation engines to reinforcement-learning locomotion platforms.
View Projects
Transformer-based RL policies for 8-DOF point-foot biped locomotion, using student-teacher distillation for sim-to-real transfer. Built on the blank engine. This is the evolution from veritas project applying the fast update rate of the neural network with the power of transformers. The multi-head attention model seems perfect for the complexity of a walking robot.